The Evolution of Retirement by J. Ignacio Conde-Ruiz* Vincenzo ...

The Evolution of Retirement
by
J. Ignacio Conde-Ruiz*
Vincenzo Galasso**
Paola Profeta***
DOCUMENTO DE TRABAJO 2005-03
February 2005
* Prime Minister’s Economic Bureau and FEDEA.
** IGIER, Universit`a Bocconi and CEPR.
*** Università di Pavia and Università Bocconi.
Los Documentos de trabajo se distribuyen gratuitamente a las Universidades e Instituciones de Investigación que lo solicitan. No obstante están
disponibles en texto completo a través de Internet: http://www.fedea.es/hojas/publicaciones.html#Documentos de Trabajo
These Working Documents are distributed free of charge to University Department and other Research Centres. They are also available through
Internet: http://www.fedea.es/hojas/publicaciones.html#Documentos de Trabajo

Depósito Legal: M-2958-2005

The Evolution of Retirement ∗
J. Ignacio Conde-Ruiz
Prime Minister’s Economic Bureau and FEDEA
Vincenzo Galasso
IGIER, Università Bocconi and CEPR
Paola Profeta,
Università di Pavia and Università Bocconi
February 2005.
Abstract
We provide a long term perspective on the individual retirement behavior
and on the future of early retirement. In a cross-country sample, we
find that total pension spending depends positively on the degree of early
retirement and on the share of elderly in the population, which increase
the proportion of retirees, but has hardly any effect on the per-capita pension
benefits. We show that in a Markovian political economic theoretical
framework, in which incentives to retire early are embedded, a political
equilibrium is characterized by an increasing sequence of social security
contribution rates converging to a steady state and early retirement. Comparative
statics suggest that aging and productivity slow-downs lead to
higher taxes and more early retirement. However, when income effects
are factored in, the model suggests that periods of stagnation — characterized
by decreasing labor income — may lead middle aged individuals to
postpone retirement.
Keywords: pensions, lifetime income effect, tax burden, politicoeconomic
Markovian equilibrium
JEL Classification: H53, H55, D72
∗ Paola Profeta, paola.profeta@unibocconi.it. Jose Ignacio Conde-Ruiz, jiconde@presidencia.gob.es;
Vincenzo Galasso, vincenzo.galasso@unibocconi.it. We thank
J.F.Jimeno for many useful comments and suggestions, G.Casamatta, P. Poutvaara and
participants to the Fourth CEPR/RTN workshop “Financing Retirement in Europe” at
CORE, Louvain-la-Neuve and to the 2004 Annual Meeting of the IIPF. All remaining errors
are ours. Part of this paper was written while P.Profeta was research fellow at CORE,
Université Catholique de Louvain, under the CEPR-RTN Project “Financing Retirement
in Europe”. Financial support from Fundación Ramon Areces and Fundación BBVA is
gratefully acknowledged. The views expressed herein are those of the authors and not
necessarily those of the Spanish Prime Minister’s Economic Bureau.
1

1 Introduction
Retirement decisions represent one of the hottest issues of the current social
security debate. Several studies - see Blondal and Scarpetta (1998) and Gruber
and Wise (1999 and 2003) among the most recent work - have suggested that
individual retirement decisions are strongly affected by the design of the social
security system. In particular, individuals tend to retire either as soon as they
are given the opportunity, i.e., at early retirement age, or at normal retirement
age. Moreover, most social security systems have been proven to provide strong
incentives - in terms of large implicit taxes on continuing to work - to anticipate
retirement. In taking their retirement decisions, most individuals prefer to enjoy
generous early retirement benefits - and the leisure associated with an early exit
from the labor market - rather then to continue working, since, in the latter
case, their additional contributions to the system would not sufficiently increase
their future pension benefits.
Several studies have made an additional step by arguing that the massive
use of early retirement provisions has come at a cost: the deterioration of the
financial sustainability of the system, already under stress because of population
aging. In fact, several international organizations - such as the European
Union at the 2001 Lisbon Meetings - have advocated an increase in the effective
retirement age, or - analogously - the increase in the activity rate among individuals
aged above 55 years, as a key policy measure to control the rise in social
security expenditure. In a nutshell, the postponement of the retirement age has
become a milestone in all social security reform’s proposals. However, whether
these policy prescriptions will actually be adopted depends on the politics of
early retirement (see Cremer and Pestieau 2000, Fenge and Pestieau, 2005, for
a detailed discussion of early retirement issues, and Galasso and Profeta, 2002,
for a more general survey of the political economy of social security).
2

In this paper, we aim at providing a long term perspective on the individual
retirement behavior. We start from the role played at present by retirement
decisions in social security and we analyze implications for the future of early
retirement.
First, we address the relation between the use of the early retirement provisions
and the pension expenditure in a cross-country sample to evaluate the
impact of the effective retirement age on the level of pension expenditure. We
find that the total size of pension expenditure depends positively on different
measures of early retirement, on the share of elderly in the population and on
per capita GDP. To better investigate these relations, we decompose the total
pension expenditure over GDP in the number of pensioners over total population
and the average pension over per capita GDP. We find that early retirement
shows a positive relation with the number of elderly, but hardly any relation
with the average pension.
Second, we provide a simple theoretical framework in which the incentives to
retire early - as identified by Blondal and Scarpetta (1998) and Gruber and Wise
(1999 and 2003) - are examined in conjunction with some long term determinants
of the retirement decision.
We use a Markovian politico-economic model to
predict the equilibrium path of fiscal policies over social security. We find a
political equilibrium characterized by an increasing sequence of social security
tax rates converging to a steady state and by a parallel increment in the use
of early retirement provisions. Comparative statics suggest that aging will lead
to higher taxes and more early retirement. However, when income effects are
factored in, the model suggests that period of stagnation — characterized by
decreasing labor income — may lead middle aged individuals to increase their
labor supply and hence reduce early retirement.
There exists a vast literature on retirement decisions. Already two decades
ago, Feldstein (1974) and Boskin (1977) analyzing the determinants of the de-
3

cline in the labor force participation of elderly workers pointed at two key parameters
of social security systems: the income guarantee and the implicit tax
on earnings.
Endogenous retirement decisions have been analyzed by showing
how pensions systems introduce distortions in the labor supply choice (see
among others Diamond and Mirrless, 1978, Hu, 1979, Crawford and Lilien,
1981, and Michel and Pestieau, 1999). The empirical literature of the effects
of social security design on retirement includes Boskin and Hurd (1978), Hurd
and Boskin (1984), Diamond and Housman (1984), and more recently Gruber
and Wise (1999 and 2003) and Blondal and Scarpetta (1998). A new literature
has lately emerged on the political economy of early retirement (see Fenge and
Pestieau, 2004, Lacomba and Lagos, 2000, Casamatta et al., 2002, Conde-Ruiz
and Galasso, 2003 and 2004, Cremer and Pestieau 2003), although generally neglecting
the role of income effects. Markovian politico-economic models of social
security have been recently studied by Azariadis and Galasso (2002), Hassler et
al (2003), Gonzalez-Eiras and Niepelt (2004) and Forni (2005).
The paper is structured as follows.
In the next section, we address the
relation between retirement decisions and social security expenditure in a crosscountry
sample. Section 3 presents a Markovian politico economic model, in
which individual retirement decisions and the evolution of social security and
early retirement are analyzed. Section 4 concludes.
2 Some Evidence on Retirement and Pension
Expenditures
The relation between retirement and social security has been largely analyzed
(see Latulippe, 1997, and Gruber and Wise, 1999 and 2003, among many others).
The social security system may induce early exit from the labor market before
collecting pension benefits or may provide large incentives to retire early. It
is however worth noticing that the relation between retirement behavior and
4

social security expenditure seems to depend crucially on the number of elderly
in the population (see Profeta 2002). In fact, in countries with a large share of
elderly in the population, the effective labor force participation of the elderly
tends to be low. These countries are also associated with a large total pension
expenditure.
What are then the main determinants of social security expenditures? And
in particular, what is the role of the retirement behavior?
To answer these
questions, we build a large data set collecting information on demographics,
retirement and social security for many countries around the world and perform
a simple cross-country analysis.
Asdemographicvariableweusetheproportionofelderly(individualsaged
more than 65 years) in the total population (OLD), taken from the United Nations
Demographics Yearbook (1999). From the Yearbook of Labor Statistics
of the International Labour Office (1985-2000) we construct two variables that
measure the effective retirement age: (i) the average activity rate in the period
1985-2000 for individuals between 55 and 64 years old (ACTOT for all and
ACTOTM for man) and (ii) the difference between the average activity rate
in the period 1985-2000 for individuals between 40 and 55 years old (AC1TOT
for all and AC1M for man) and the average activity rates for individuals between
55 and 64 years old (DAC for all and DACM for man). This last variable
(DAC) represents a measure of the extent of early retirement, by capturing the
drop in the activity rate from age 40-55 to age 55-64. From the International
Monetary Fund Government Finance Statistics Yearbook (several years in the
period 1986-1998) we obtain the average level in the period 1986-1998 of pension
expenditures as percentage of GDP (PENSEXP) 1 .Finally,fromtheILO
World Labour Report (2000) we obtain the percentage of pensioners over total
1 Notice that the IMF’s definition of social security includes old age payments, and therefore
it differs from an ideal measure of “governments transfers to the elderly”, because it excludes
medical and other subsidies for the elderly.
5

population (NUMPENS). Combining this information and the level of pension
expenditures as percentage of GDP, we calculate the per capita average pension
(AVPENS).
Table 1 provides some simple statistics of our variables.
We perform a simple cross-country regression. Table 2 summarizes the results
for a sample of all countries for which the data are available 2 and table
3 for a sub-sample of OECD countries 3 . The level of pension expenditures as
percentage of GDP depends significantly and positively on the number of elderly
andonGDP,andsignificantly but negatively on the activity rate of individuals
between 55 and 64 years old (column (a)). According to this result, countries
with a larger share of elderly in their populations spend more for pensions, as
intuitively expected. Richer countries also spend more in pension, as predicted
by Wagner’s law. Finally, countries where the effective retirement age is higher
(i.e. individuals between 55 and 64 years old have higher activity rates) spend
less in pensions. These determinants together explain most of the variability in
pension expenditures (R*2=0.837). To obtain a better measure of early retirement,
we use the variable DAC, which represents the reduction of the activity
rate of individuals between 55 and 64 years old with respect to the activity rate
of people in age 40-55. This measure of early retirement is significantly and
positively related to the level of pension expenditures (column (b)), suggesting
that countries which use more early retirement spend more for pensions.
A
2 Albania, Algeria, Argentina, Australia, Austria, Bahamas, Bahrain, Bangladesh, Barbados,
Belarus, Belgium, Belize, Benin, Bolivia, Brazil, Bulgaria, Canada, Cape Verde, Central
African Republic, Chile, China, Colombia, Costa Rica, Croatia, Cyprus, Czech Republic,
Denmark, Dominica, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Fiji, Finland, France,
Germany, Greece, Grenada, Guatemala, Guyana, Hungary, Iceland, Indonesia, Iran, Ireland,
Israel, Italy, Jamaica, Japan, Jordan, Korea, Latvia, Lithuania, Luxembourg, Malaysia, Mauritius,
Mexico, Moldova, Netherlands, New Zealand, Nicaragua, Nigeria, Norway, Pakistan,
Panama, Poland, Portugal, Romania, Senegal, Singapore, Slovakia, Spain, Sri Lanka, Sweden,
Switzerland, Trinidad and Tobago, Tunisia, Turkey, Ukraine, United Kingdom, United States,
Uruguay.
3 Australia, Austria, Belgium, Canada, Czeck Republic, Denmark, Finland, France, Germany,
Iceland, Ireland, Italy, Japan, Korea, Luxembourg, Mexico, Netherlands, New Zealand,
Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States.
6

similar result is obtained if we restrict to the activity rates of man (column (c)).
To better understand the determinants of the level of pension expenditures
and the role played by each of the identified variables, as pointed out by Jimeno
(2002a, 2002b) we use the following identity:
PENSEXP = NUMPENS ∗ AV P ENS
which makes clear how the level of pension expenditures as percentage of GDP is
the product of the number of pensioners as a proportion of total population and
of the average pension in terms of GDP per capita. According to this decomposition,
the role played by the demographic variable (the proportion of elderly in
the population) and the retirement variables (the measures of early retirement)
in explaining the level of pension expenditures can be better understand using
two additional sets of regressions. Due to limited information on the number
of pensioners and the average pension, we perform these regressions only using
thesampleofallcountries.
First, we try to estimate the determinants of the number of pensioners,
as shown in table 4.
Clearly, countries where there are more elderly in the
population have more pensioners (OLD is positively and significantly related
to NUMPENS). Countries where there is more early retirement have also more
pensioners (ACTOT is negatively and significantly related to PENSEXP in
column (a) and DAC negatively and significantly in column (b)). This regression
makes clear that both the proportion of elderly in total population and early
retirement increase the level of pension expenditures through the increase of
the number of pensioners. Second, we estimate the impact of demographics and
retirement on the level of average pension. Table 5 shows that neither aging
(OLD) nor early retirement (DAC) are significant in explaining the average
pension.
To summarize, these simple results suggest that while more elderly and lower
7

early retirement increase the overall pension expenditures, via asignificant increase
of the number of retirees, they do not affect the per capita average pension.
In the following section we introduce a political economy model which is
able to account for these facts.
3 A politico-economic model
3.1 The economic Environment
We consider a simple three -period overlapping generations model. Every period
three generations are alive, we call them respectively young, adult and old.
Young face a labor-leisure trade off on the intensive margin. They choose how
many hours of work to supply, receive a wage, w y , pay a payroll tax on labour
income, τ , and save all their disposable income for old age consumption. There
exists a storage technology that transforms a unit of today’s consumption into
1+r units of tomorrow’s consumption. All private intertemporal transfers of
resources into the future are assumed to take place through this technology.
Adult individuals decide what fraction, z, of the second period to spend
working; in other words, they decide when to retire. An adult individual who
works a proportion z of the second period receives a net labor income equal
to w a (1 − τ), for the fraction z of the period.
To simplify the analysis, we
assume that no pension benefit is received for the remaining fraction (1 − z),
during which individuals however enjoy leisure. Finally, all old individuals are
assumed to retire and to obtain a pension, p. Population grows at a rate n.
The life time budget constraint for an agent born at time t is then equal to:
c o t+2 =(1− τ t ) l t w y t (1 + r) 2 +(1− τ t+1 ) z t+1 w a t+1 (1 + r)+p t+2 (1)
where c o t+2 is old age consumption at time t+2, l t represents the amount of work
supplied by a young individual at time t and subscripts indicate the calendar
time. Moreover, τ t and τ t+1 are the pay roll taxes respectively at periods t and
8

t +1andr is the interest rate.
To account for differences in the wages by age, i.e., between young and adult
workers, and for economic growth, we characterize the relations between wages
across time and generations as follows: wt+1 a =(1+g) wt a ,whereg is the real
growth rate of wages, and w y t
= ϕw a t ,whereϕ defines the ratio between the
wage of young and adult workers at time t.
To simplify the analysis, we assume that agents only value life-time income
— or analogously old age consumption, c t+2 — and leisure when young and adult,
according to the following utility function, which has been introduced in a two
periods context by Casamatta, Cremer and Pestieau (2002) and Cremer and
Pestieau (2002) to study incentive effects on early retirement provisions:
U (l t ,z t+1 ,c t+2 )=c o t+2 − l2 t
2α wy t − z2 t+1
2γ wa t+1 (2)
where the second and third terms represent the disutility from the effective
labor supply, and α and γ are parameters that quantify the relative importance
of leisure when young and adult, and that we take to be equal respectively to
α =1/ (1 + r) 2 and γ =1/ (1 + r). 4
Hence, a young agent at time t and an adult at time t+1 maximize eq. 2 with
respect to l t and z t+1 subject to the budget constraints at eq. 1. The solution
of the two maximization problems yields the following optimal individual labor
supply decisions:
b lt = 1− τ t (3)
bz t+1 = 1− τ t+1 (4)
We consider a balanced budget pay as you go (PAYG) social security system,
where the sum of all pension transfers is equal the sum of all contributions. Since
pensions are awarded to elderly individuals only, whereas young and adults
4 These assumptions simplify the analysis. However the main results qualitatively hold true
in more general cases.
9

workers contribute according to their labor income, the full pension transfer at
time t which balances the budget constraint is equal to:
p t = τ t
³
(1 + n t )(1+n t−1 ) w t
b lt +(1+n t−1 ) w a t bz t´
(5)
or equivalently
where w t =((1+n t ) w y t + w a t ).
p t = τ t (1 − τ t )(1+n t−1 ) w t (6)
Finally, by substituting the individual decisions at eq.
3 and 4 and the
above social security budget constraint, we derive the following expression for
the indirect utility function of an adult individual at time t:
V a (τ t−1 ,τ t ,τ t+1 ) = (1− τ t−1 ) b l t−1 w y t−1 (1 + r)2 +(1− τ t ) bz t w a t (1 + r)+
3.2 The Political Equilibrium
+τ t+1 (1 − τ t+1 )(1+n t ) w t+1 − b l 2 t−1
2α wy t−1 − bz2 t
2γ wa t .(7)
The purpose of this paper is to propose a theoretical framework in which to account
for the link between early retirement provision and the size of the social
security system examined in section 2. We have already showed at eq. 4 that
early retirement behavior may be induced by specific featuresofthesocialsecurity
system, such as the contribution rate. Here, we analyze the determination
of this social security contribution rate within the political arena. We follow
a well established tradition in political economics by concentrating on the median
voter decision. Yet, due to the intergenerational nature of the system, we
allow for some interdependence between current and future political decisions.
In particular, we analyze Markov perfect equilibrium outcomes of a repeated
voting game over the social security contribution rate. Models using subgame
perfection as equilibrium concept identify a large set of sequences of equilibrium
social security contribution rates that need not to depend on the state on the
10

economy, while in models that concentrate on Markov perfect equilibrium outcomes,
the political decision over the social security contribution rate depends
on the state of the economy — typically through the stock of capital 5 . Since
the aim of the paper is to examine the possible link between the use of early
retirement provisions and the size of the social security system, we choose to
summarize the state of the economy for our Markov equilibrium by the share of
early retirees in the population. This amounts to assume that economies with
the same share of early retirees take the same decision on the size of the system.
More specifically, at every period t, the median voter in each generation of
voters — hence typically an adult individual — decides her most favorite social
security system (i.e., the tax rate τ t ). In taking her decision, she expects her
current decision to have an impact of future policies. In particular, her expectations
about the future social security tax rate — and hence about her pension
benefits — depend on the current level of early retirement, according to a function
τ t+1 = q e (z t ). Hence, future contribution rates depend on the current level
of labor force participation by the elderly, which is in turn affected by the current
voter’s decision over the social security contribution rate. Therefore, the
median voter´s optimal decision can be obtained maximizing her lifecycle utility
with respect to τ t and given expectations on the next period policy function
τ t+1 = q e (z t )=Q(z t (τ t )):
max V a (τ t−1 ,τ t ,τ t+1 )=maxV a (τ t−1 ,τ t ,Q(z t (τ t ))) (8)
τ t
τ t
We can now define the Markov political equilibrium as follows
Definition 1 A Markov political equilibrium is a pair of functions (Q, Z), where
Q :[0, 1] → [0, 1] is a policy rule, τ t = Q(z t−1 ),andZ :[0, 1] → [0, 1] is a private
5 For examples of Markov equilibria, see Krusell et al.(1997), Grossman and Helpman
(1998), Bassetto (1999), Azariadis and Galasso (2002), Hassler et al. (2003), Gonzalez-Eiras
and Niepelt (2004), Forni (2005).
11

decision rule, z t =1− τ t = Z(τ t ), such that the following functional equations
hold:
i) Q(z t−1 ) = arg max
τ t
V a (τ t−1 ,τ t ,τ t+1 ) subject to τ t+1 = Q(Z(τ t )).
ii) Z(τ t )=1− τ t
The first equilibrium condition requires that τ t maximizes the objective function
of the median voter — an adult individual as long as n t (1 + n t−1 ) < 1—
taking into account that the future social security system tax rate, τ t+1 depends
on the current social security tax rate, τ t , via the private labor supply
decision of the adults. Furthermore, it requires Q(z t−1 )tobeafixed point in
the functional equation in part i) of the definition. In other words, if agents
believe future benefits to be set according to τ t+j = Q(z t+j−1 ), then the same
function Q(z t−1 )hastodefine the optimal voting decision today. The second
equilibrium condition requires that all old individuals choose their labor supply
optimally.
In order to compute the Markov political equilibrium, we have to consider
the optimal social security tax rate chosen by the median voter at time t, who
maximizes the indirect utility function at eq. 7 with respect to τ t and subject
to τ t+1 = Q(Z(τ t )).
The corresponding first order condition is:
−w a t z t (1 + r)+ dp t+1
dτ t+1
Q 0 (z t )Z 0 (τ t )=0 (9)
where Z 0 (τ t )=−1, which suggests that the current cost of contributing to the
system has to be compensated by a corresponding benefit in the future. The
solution of the maximization problem of the median voter yields the optimal
fiscal policies, as summarized in the following proposition.
Proposition 2 The set of feasible fiscal policies {τ ∗ t } ∞ t=s
∈ [0, 1] which can be
12

supported by a Markovian politico-economic equilibrium satisfies:
s
τ t+1 = Q(Z(τ t )) = 1 2 − 1 1 − 2A − 2(1 + r)wa t Z(τ t ) 2
2
(1 + n t )w t+1
where A, the free parameter pinned down by the first median voter’s expectation
of future policies, is restricted to the support A ∈ [(1 + r)w a t , (1 + n t )w t+1 /2].
Proof. See Appendix.
The result in proposition 2 points to the existence of a positive link between
thecurrentuseofearlyretirementprovisions and the future social security
contribution rate. This link completes the economic chanel running from the
social security contribution rate to the current labor supply decision of the
adults, as described at eq. 4. In particular, a current increase in the social
security contribution rate leads to more current early retirement — by reducing
the opportunity cost of retirement — which in turn creates expectations for more
social security contributions — and hence more early retirement — in the future.
In fact, since z t =1− τ t , the dynamics for the equilibrium policy function
can be described as follows:
τ t+1 = 1 2 − 1 2
s
1 − 2A − 2(1 + r)wa t (1 − τ t ) 2
(1 + n t )w t+1
(10)
Furthermore, it is easy to show that dτ t+1 /dτ t
∈ (0, 1), and thus the equilibrium
path converges to a stable steady state with a positive social security
contribution rate — and early retirement. As shown in figure 1, if the initial
social security contribution is below the steady state level, the dynamics feature
an increasing sequence of tax rate (and an increasing use of early retirement),
which converges to the steady state.
3.3 Aging, Social Security and Early Retirement
The equilibrium policy function obtained in the previous section allows us to
analyze the effects of aging on the dynamics of the social security tax rate and
13

on the use of early retirement.
To see how aging may affect this evolution,
consider the decision of the median voter at time t.
When determining the
current social security contribution rate, this adult voter follows the equilibrium
function described at eq. 10 — lagged one period. By setting the policy according
to this function, she validates the expectations of the previous median voter,
since she is indifferent on what to vote, provided that she expects the next
median voter to follow the same function in the future.
Consider now that at time t information becomes available that aging will
occur at time t + 1. In other words, at time t, the median voter learns that
population growth will permanently drop from the following period: n t+i 0. Since no change occurs in the policy function at time t —noticefrom
eq. 10 that τ t = Q(Z(τ t−1 )) only depends on n t−1 and on n t —thesocial
security contribution rate will be set in accordance with the expectations of the
median voter at time t − 1. Yet, the current median voter’s expectations on the
future social security contribution rate do change because of the expected aging
and so will the policy function in the next and in all future periods. The next
proposition summarizes the results of this expected aging process on the social
security contribution rate and on early retirement.
Proposition 3 Consider an economy at its stable steady state characterized by
the social security contribution τ, a pension transfer p and by z =1− τ. A
permanent decrease in the population growth rate taking place at time t +1,but
anticipated by the voters at time t — that is, n t+i 0, withE t (n t+i )=
n t+i — has the following effects:
• at time t +1, it increases the contribution rate, τ t+1 >τ t ,andtheearly
retirement, z t+1
< z t , while leaving the per-capita pension unaffected,
p t+1 = p t ;and
• at steady state, it increases the social security contribution rate — i.e, τ 0 >
14

τ — the use of early retirement — i.e., z 0 < z.
Proof. See Appendix.
According to the proposition above, aging leads to higher social security tax
rates and more early retirement, while leaving the per-capita pension transfer
unaffected, at least in the short run.
The intuition behind this result is the
following. As the population growth rate drops, the implicit return from a PAYG
social security system decreases as well. Hence, the median voter modifies the
policy function by making it more responsive to the past early retirement (and
hence to the past social security contribution rates) in order to compensate the
lower returns with higher contributions. In particular, the median voter at time
t + 1 — that is, when the aging process effectively begins — increases the tax rate
to counterbalance the effect of the aging on the pension transfer to the current
elderly, which in fact remains constant. In future periods, until a new steady
state is reached, contribution rates will increase to finance higher social security
spending and to compensate the successive median voters — the adults — for the
higher contributions paid on their labor income. Eventually, at the new steady
state, the contribution rate increases leading to more early retirement.
Another component of the returns from social security that has received
large attention in the literature is the growth rate of wages, g. In fact, higher
economic growth is often advocated as a painless solution to the financial sustainability
problems affecting most PAYG social security systems.
Yet, the
political economy literature on social security has pointed out that changes in
economic growth — by affecting the social security returns — modify the incentives
faced by the voters in deciding the future pension policy.
To see how higher economic growth may affect these dynamics, consider the
decision of the median voter, who determines the social security contribution
rate, at time t, when a permanent increase in the economic growth rate takes
15

place, g t+i >g t−1 ∀i ≥ 0. In our Markov perfect equilibrium, the median voter
does not change her voting behavior, in order to validate the expectations of
the previous median voter. Yet, she expects future median voters to obey to
the new policy function — see eq.
10 — where the new values of the growth
rates are taken into account. The next proposition summarizes the results of
this increase in economic growth on the social security contribution rate and on
early retirement.
Proposition 4 Consider an economy at its stable steady state characterized by
the social security contribution τ and by z =1− τ. A permanent increase in
theeconomicgrowthrateattimet —thatis,g t+i >g t−1 ∀i ≥ 0 —decreasesthe
steady state social security contribution rate — i.e, τ 0 < τ —andreducestheuse
of early retirement — i.e., z 0 > z.
Proof. See Appendix.
Our Markov political equilibrium hence confirms the conventional (economic)
wisdom that more growth reduces the size of the social security system. The
intuition is similar, yet specular, to the case of aging. Higher growth now rises
the implicit returns from a PAYG system. Hence, the pension benefits to the
retirees may be obtained with lower contribution rates, and the policy function
becomes less responsive to the past early retirement (and hence to the past
social security contribution rates).
Eventually, at the new steady state, the
contributionratedropsleadingtoaloweruse of the early retirement provisions.
3.4 Social Security, Early Retirement and Income Effects
The results obtained in the previous section are hardly reassuring. According to
our political economy analysis, aging and productivity slow-downs will lead to
larger social security system and more early retirement. However, the analysis
carried out in the previous sections did not incorporate the possible effect of
16

changes in income or wealth on the individual early retirement decisions, and
thus on the politically determined social security contribution rate. According
to eq. 4, in fact, retirement decisions only depended on substitution effects —
through the impact of the labor tax on continuing to work. Possible income
effects — leading poorer individuals to work longer, i.e., to retire later — were
abstracted from, although several authors (see for instance Costa, 1998) have
suggested that the long lasting decreasingtrendintheretirementagemayat
least partially be due to the major improvements in economic conditions that
increased the demand for leisure, and hence for early retirement.
In this section, we introduce a simple income effect in our economic model.
In particular, we modify the utility function at eq. 2 by adding a new term for
the disutility of labor which depends on the adult’s wage:
U (l t ,z t+1 ,c t+2 )=c o t+2 − l2 t
2α wy t − z2 t+1
2γ wa t+1 − λz t+1
¡
w
a
t+1
¢ 2
(11)
This modification 6 of the previous specification amounts to assume that the
adults’ retirement decision depends negatively on their current income. In fact,
their optimal individual labor supply decision becomes:
bz t+1 =1− τ t+1 − δw a t+1 (12)
where δ = λ∗γ, and the growth rate of wages, g, is assumed to be equal to zero.
Thus, low-income adults decide to work for longer periods and richer adult to
retire earlier. This result suggests that, unlike in Cremer and Pestieau (2003),
theincomeeffect may dominate the substitution effect 7 .
Clearly, this change in the retirement behavior of adults affects the PAYG
social security systems, since the full pension transfer at time t which balances
6 An alternative, more satisfactory way of introducing an income effect in the retirement
decision would be to simply eliminate the linearity from the utility function. Were the utility
function to be concave in the level of consumption, the retirement decision would immediately
depend (negatively) on the lifetime income of the agent. However, this alternative specification
would not yield a close form solution for the Markov perfect political economic equilibrium.
7 See Jensen et al. (2004).
17

the budget constraint becomes:
h
p t = τ t (1 + n t−1 ) w t (1 − τ t ) − δ (wt a ) 2i . (13)
How does the introduction of this negative income effect in the retirement decision
modify the result of the Markov political equilibrium described in section
3? In this new scenario, the sequence of median voters face a similar political
decisions to the one described in section 3.2, which leads to the following
dynamics for the equilibrium social security contribution rates:
Ã
τ t+1 = 1 1 − δ ¡ ¢
wt+1
a 2
!
+ (14)
2 w t+1
v
− 1 u
Ã
t
1 − δ ¡ ¢
wt+1
a 2
! 2
− 2[A − (1 + r)wa t (1 − τ t − δwt a ) 2 ]
.
2 w t+1 (1 + n t )w t+1
³
with A ∈ [(1 + r)wt a (1 − δwt a ) , (1 + n t ) w t+1 − δ (wt a ) 2´2 /2wt+1 ].
Hence, also in this scenario, the dynamics described at eq. 14 points to the
existence of a positive link between the current use of early retirement provisions
and the future social security contribution rate. A higher current social security
contribution rate induces more current early retirement — due to the reduction
in the opportunity cost of retiring — and creates expectations for more social
security contributions and early retirement in the future. Moreover, as in the
previous case, it is easy to show that dτ t+1 /dτ t ∈ (0, 1), and thus the equilibrium
path converges to a steady state with a positive social security contribution rate
— and early retirement, with equilibrium paths that are qualitatively similar to
the ones shown in figure 1.
The discussion above suggests that this scenario, which incorporates an income
effect, features similar properties for the Markov equilibrium outcomes
as in the previous case. However, this new specification allows an analysis of
the impact of a reduction in the adult’s labor income — i.e., a negative income
effect — on the adults’ retirement behavior, in which the effect on the sequence
18

of equilibrium social security contribution rates is also taken into account.
To see how this negative income effect may modify these dynamics, we consider
an unexpected and permanent reduction in the labor income of the adults
(and of the young) that takes place at time t, wt+i a z
and z t+1 > z.
Proof. See Appendix.
This proposition provides an interesting insight on the future of the early
retirement provisions, which complements the results obtained in the previous
section.
When the effects of changes in income or wealth on the retirement
behavior are taken into account, a reduction in the adult labor income induces
individuals to postpone retirement, although we are not able to evaluate the
impact on the equilibrium social security contributions. We argue that — to the
extent that this reduction in the adults’ labor income may proxy for a drop
in the life-time labor income — this may prove a crucial result
to understand
the future evolution of the early retirement provision. Societies characterized
by economic stagnation or raise in inequalitythatincreasetheshareoflow-
19

income individuals may thus be associated with a less pervasive use of these
early retirement provisions.
4 Conclusions
Since the recent studies by Blondal and Scarpetta (1998) and Gruber and Wise
(1999 and 2003), among others, providing evidence that individual retirement
decisions are strongly affected by the design of the social security system, measures
to postpone the effective retirement age has become a milestone in all
social security reform’s proposals.
We concentrate on the current and future role of retirement, by analyzing the
individual retirement decisions under different perspectives. First, we analyse
the relation between the use of the early retirement provisions and the social
security expenditure. In a cross-country sample, we find that the level of pension
expenditures depends positively and significantly on the proportion of elderly
in the population (demographic variable) and on measures of early retirement
(retirement variables). This total effect can be decomposed in the effect on the
number of pensioners as proportion of total population and the effect on the
per capita average pension. An older population and/or more early retirement
imply more pensioners, while it has hardly any effect on the generosity of the
system, i.e., on the per capita average pension.
We then introduce a political economy theoretical model to analyse the long
term determinants of early retirement and the evolution of the social security
system and of the early retirement provisions. In our politico-economic Markovian
environment, every period an adult median voter determines the social
security contribution by considering the evolution of the early retirement behavior.
We use two specifications of our economic environment. In the former model,
20

we emphaze the relevance of the incentives to retire early — i.e., the substitution
effect — in line with a large empirical literature which shows how (at the margin)
non-actuarially fair pension systems may induce rational agents to retire
early, by reducing the opportunity cost of leisure. In this scenario, we obtain
a political equilibrium characterized by a raising sequence of social security tax
rates converging to a steady state. Along this equilibrium path, the use of early
retirement provisions is increasing. The results of the comparative statics on
this specification are dismaying. In line with the conventional wisdom in the
economic literature, but unlike most of the implications found in the political
economy literature (see Galasso and Profeta, 2002), productivity slow-downs
(and aging) are expected to lead to higher social security contributions and to
more early retirement.
In the latter economic specification, we introduce a negative effect of income
on the retirement decisions: a decrease in the labor income when adult leads to
postpone retirement. In this scenario, which aims at including an additional long
term determinant of the retirement decision — namely the life-time income — the
political economic equilibrium is still characterized by an increasing sequence
of social security tax rates and early retirement converging to a steady state.
Yet, comparative statics now suggest that a decrease in the adult labor income
will reduce the use of early retirement provisions, even after the effects on the
equilibrium social security contributions are considered. To the extent that this
change in adult income may proxy for a change in the net life-time income,
we believe that this may prove an important result for the evolution of the
early retirement provisions. In fact, we notice that, by reducing the internal
rate of return to social security, aging also decreases the life-time net income of
the new generations — the more so, the more the system is used. In its initial
specification — with no income effects — our politico-economic model suggests
that aging leads to a large social security system and to more early retirement.
21

Yet, to the extent that this also translates into a lower life-time income, an
additional channel may arise that reduces the use of early retirement provisions
and hence postpones retirement.
References
[1] Azariadis, C. and Galasso, V., 2002. “Fiscal Constitutions”. Journal of
Economic Theory 103, 255-281.
[2] Bassetto, M. , 1999. “Political Economy of Taxation in an Overlapping-
Generations Economy”, Federal Reserve Bank of Minneapolis Discussion
Paper, no. 133
[3] Boskin, M. 1977. “Social Security and Retirement Decisions.” Economic
Inquiry, Vol. 15 (1), 1-25.
[4] Boskin and Hurd, 1978. “The effect of social security on early retirement”.
Journal of Public Economics, 10, 361-377.
[5] Blondal S. and S. Scarpetta, 1998. “Falling Participation Rates Among
OlderWorkersinOECDCountries:theRoleofSocialSecuritySystems”,
OECD Economic Department Working Paper.
[6] Casamatta, G., H. Cremer and P. Pestieau, 2002. “Voting on Pensions with
Endogenous Retirement Age”, mimeo.
[7] Costa, D.L., 1998, “The Evolution of Retirement: An American Economic
History 1880-1990”, University of Chicago Press.
[8] Conde-Ruiz J.I. and V. Galasso, 2003. “Early Retirement”, Review of Economic
Dynamics, 6, 12-36.
[9] Conde-Ruiz J.I. and V. Galasso, 2004, “The Macroeconomics of Early Retirement”,
Journal of Public Economics 88 (9-10):1849-1969.
22

[10] Costa, D. (1998). “The Evolution of Retirement”. Chicago: University of
Chicago Press.
[11] Crawford V.P. and D. Lilien, 1981. “Social Security and retirement decision”,
Quarterly Journal of Economics, 95, 505-529.
[12] Cremer, H. and Pestieau, P., 2003, “The Double Dividend of Postponing
Retirement”, International Tax and Public Finance 10, 419-434.
[13] Cremer, H., J. Lozachemeur and P. Pestieau, 2002, “Social Security, Retirement
Age and Optimal Income Taxation”. mimeo
[14] Cremer, H., and P. Pestieau, 2000, “Reforming Our Pension System: Is
it a Demographic, Financial or Political Problem?”, European Economic
Review, 40, 974-983.
[15] Diamond P. and J. Housman, 1984 “The retirement and Unemployment
Behavior of older men” in Aaron, H. and Burtless, G (eds). Retirement and
economic behaviour (Washington, D.C. Brookings Institution), 97-134.
[16] Diamond, P. A. and J. A. Mirrless “A model of social Insurance and with
variable retirement”, Journal of Public Economics 10, 294-336
[17] Fenge R. and Pestieau, P.,“Social security and retirement” MIT Press
forthcoming
[18] Feldstein, M., 1974. “Social Security, induced retirement, and Aggregate
Capital Accumulation”, Journal of Political Economy, 82(5), 905-926.
[19] Forni, L., 2005 “Social security as Markov equilibrium in OLG Models”
Review of Economic Dynamics 1, 178-194.
[20] Galasso, V., and P. Profeta, 2002, “The Political Economy of Social Security:
A Survey”, European Journal of Political Economy, 18, 1-29.
23

[21] Gonzalez-Eiras, M. and D. Niepelt, 2004. “Sustaining Social Security”. WP
371, IIES Stockholm.
[22] Grossman, G. and Helpman, E. 1998. “Intergenerational redistribution with
short-lived governments” 108: 1299-1329.
[23] Gruber, J. and D. Wise (eds.), 1999. “Social Security and Retirement
Around the World”, University of Chicago Press, Chicago.
[24] Gruber, J. and D. Wise (eds.), 2004, “Social Security Programs and Retirement
Around the World: Micro Estimation”, University of Chicago Press.
[25] Hassler, J. Rodriguez-Mora, J.V., Storesletten, K. and Zilibotti, F., 2003,
“The Survival of the Welfare State”, American Economic Review 93(1),
87-112.
[26] Hu, S., 1979. “Social security, the supply of labor and capital accumulation”,
American Economic Review 69 (3), 274-283.
[27] Hurd, M. and Boskin, M.J., 1984. “The Effect of Social Security on Retirement
in the Early 1970s.” Quarterly Journal of Economics, 99, 767-790.
[28] International Labour Office, 1985-2000. Yearbook of Labour Statistics.
[29] International Labour Office, 2000. World Labour Report 2000.
[30] International Monetary Fund, Government Finance Statistics Yearbook
(1985-1998).
[31] Jensen, S., Lau, M. and Poutvaara, P., 2004, “Efficiency and Equity Aspects
of Alternative Social Security Rules”, Finanzarchiv
[32] Jimeno, J. F., 2002a. “Demografía, empleo, salarios y pensiones”, Documento
de Trabajo 2002-04. FEDEA, Madrid.
24

[33] Jimeno, J. F., 2002b, “Incentivos y desigualdad en el sistema español de
pensiones contributivas de jubilación”. Documento de Trabajo 2002-13.
FEDEA, Madrid.
[34] Krusell, P. Quadrini, V. and Ríos-Rull, 1997, J.V. “Politico-Economic Equilibrium
and Economic Growth” Journal of Economic Dynamics and Control,
21:1.
[35] Lacomba, J.A: and F.M. Lagos, 2000, “Election on Retirement age”,
mimeo.
[36] Latulippe, D., 1997. “Effective Retirement Age and Duration of Retirement
in the Industrialised Countries Between 1950 and 1990”. ILO Discussion
Paper 2.
[37] Michel, P. and Pestieau, P., 1999. “Social security and early retirement in
an overlapping-generations growth model”, CORE Discussion Paper 9951.
[38] Profeta, P., 2002. “Aging and Retirement: Evidence Across Countries”,
International Tax and Public Finance 9, 651-672.
[39] United Nations, Demographic Yearbook (1990-2000).
25

5 Appendix
5.1 Proof of proposition 2
The first order condition of the median voter is:
−z t w a t (1 + r)+ ∂p t+1
∂τ t+1
Q 0 (z t ) Z 0 (τ t )=0
where
∂p t+1
=(1+n t )(1− 2τ t+1 ) w t+1 and Z 0 (τ t )=−1
∂τ t+1
Thus, the first order condition becomes
−z t w a t (1 + r) − (1 + n t ) w t+1 Q 0 (z t )+2(1+n t ) w t+1 Q 0 (z t ) Q (z t )=0
Integrating the above equation with respect to z t ,weobtain
A − z 2 t w a t (1 + r) − 2(1+n t ) w t+1 Q +2(1+n t ) w t+1 Q 2 =0
where A is a constant of integration. Solving the equation w.r.t. Q, wehave
s
Q (τ t )= 1 2 ∓ 1 1 − 2A − 2(1 + r)wa t (1 − τ t ) 2
.
2
(1 + n t )w t+1
Since τ t+1 = Q (τ t ) represents a tax rate, it has to be that Q ∈ [0, 1]. Thus, we
have that
τ t+1 = Q (τ t )= 1 2 − 1 2
s
1 − 2A − 2(1 + r)wa t (1 − τ t ) 2
(1 + n t ) w t+1
and the following restrictions on the free parameter A:
(1 + r)w a t (1 − τ t ) 2 ≤ A ≤ (1 + n t) w t+1
2
+(1+r)w a t (1 − τ t ) 2 .
Notice that, since the lower constraint is maximized for τ t+1 = 0, while the
upper constraint is minimized for for τ t+1 =1,asufficient condition for Q ∈
[0, 1] is
A ∈ [(1 + r)wt a , (1 + n t) w t+1
].
2
26

Finally, to make sure that the upper bound is indeed larger that the lower
bound, we need to assume that (1 + n t ) w t+1 > 2(1 + r)wt a ; or, analogously,
that [(1 + n t−1 ) ϕ +1]/2 > (1 + r) / [(1 + n t )(1+g)]. Q.E.D.
5.2 Proof of proposition 3
Consider the economy at its steady state and recall that eq. 10 provides the law
of motion of the social security contribution rate. At time t + 1, a decrease in
the population growth rate, n t+1 τ t ,and
— from eq. 4 — so does the early retirement, since z t+1 =1− τ t+1

w a t+1 =(1+g) w a t . By setting τ t+1 = τ t = τ in eq. 10 we obtain the following
value for the steady state social security contribution rate:
τ =
(1 + n)(1+g) w − (1 + r)wa
2(1 + n)(1+g) w − (1 + r) w a + (16)
q
(1 + n) 2 (1 + g) 2 w 2 − A [2(1 + n)(1+g) w − (1 + r) w a ]
−
2(1 + n)(1+g) w − (1 + r) w a (17)
Simple algebra shows that τ ∈ [0, 1/2) and that ∂τ/∂n < 0. Hence, aging
increases the social security contribution rate at steady state and — by eq. 4—
early retirement. Q.E.D.
5.3 Proof of proposition 4
Consider the steady state social security contribution rate at eq. 16. Simple
algebra shows that ∂τ/∂g < 0. Hence, productivity slow-downs increase the
social security contribution rate at steady state and — by eq. 4— early retirement.
Q.E.D.
5.4 Proof of proposition 5
At time t, when the adults’ and young’s labor income drops, the median voter
will not change the social security contribution rate from its steady state level,
in order to validate the expectations of the previous median voter at time t − 1.
She will however expect the tax rate to be changed at t + 1 due to this drop in
labor income. By eq. 12, it is thus immediate to see that z t =1− τ − δwt a > z
=1− τ − δwt−1, a sincewt a

∂τ t+1
∂w = − δ
(1 + n t+1 ) ϕ +1 + Θ
s µ
1 − δ(wa t+1) 2
2
w t+1
2
− 2[A−(1+r)wa t (1−τ t−δw a t )2 ]
(1+n t )w t+1
where
Θ =
Ã
δ
1 − δ ¡ wt+1
a
(1 + n t+1 ) ϕ +1
¢ 2
!
w t+1
− [(1 + n t+1) ϕ +1] £ A − (1 + r)w a t z t
2 ¤
(1 + n t )(w t+1 ) 2
− (1 + r)z t(z t − 2δw a t )
(1 + n t )w t+1
Notice that we cannot sign ∂τ t+1 /∂w; however, it is easy to show that a sufficient
µ
condition for Θ > 0isthatA w t+1 .
By eq. 12, we thus have that
∂z t+1
∂w
δ
=
(1 + n t+1 ) ϕ +1 − Θ
2(1+n t+1 )(w t+1 ) 2 − δ =
(1 + n t+1 ) ϕ
= −δ
(1 + n t+1 ) ϕ +1 − Θ
2(1+n t+1 )(w t+1 ) 2 < 0
since Θ > 0. Q.E.D.
29

Table 1. The variables (basic statistics)
Variable Obs Mean STD Error Min Max
OLD population over 65 years old/total population 106 8.487 4.935 2 18.2
ACTOT activity rate of individuals 55-64 110 0.488 0.139 0.19 0.87
ACM activity rate of man 55-64 110 0.69 0.159 0.29 0.93
AC1TOT activity rate of individuals 40-55 110 0.747 0.106 0.48 0.93
AC1M activity rate of man 40-55 110 0.921 0.074 0.35 1
DAC= AC1TOT-ACTOT 110 0.258 0.154 0 0.62
DACM=AC1M-ACM 110 0.231 0.16 0 0.55
PENSEXP Pension Expenditures as % of GDP 84 4.917 4.125 0 14.26
NUMPENS %of pensioners/ total population 53 10.503 10.41 0.01 29.7
AVPENS per capita average pension 50 1.049 1.94 0 11
GDP per capita GDP (U.S. $) 101 10850.12 9305.93 6 42769
Table 2. Determinants of pension expenditures (All countries)
Independent variables Dependent variable
PENSEXP (a) PENSEXP (b) PENSEXP (c)
CONSTANT 2.05 (0,91) -2.28 (0.42) -2.42 (0.42)
ACTOT -6.8(1.45)***
DAC 6.51 (1.77)***
DACM 6.57 (1.55)***
OLD 0.6 (0.05)*** 0.51 (0.07)*** 0.56 (0.057)***
GDP 0.00005 (0.00002)** 0.00006 (0.00002)** 0.00004 (0.00002)**
N. Obs 84 84 84
R**2 0.837 0.823 0.83
** significant at 5%, *** significant at 1%
Table 3. Determinants of pension expenditures (OECD countries)
Independent variables Dependent variable
PENSEXP (a) PENSEXP (b) PENSEXP (c)
CONSTANT 0.39 (2.36) -7.34 (2.02) -6.9 (1.69)
ACTOT -10.49 (2.51)***
DAC 11.33 (3.42)***
DACM 12.26 (2.54)***
OLD 0.76 (0.12)*** 0.7 (0.13)*** 0.69 (0.11)***
GDP 0.0001 (0.00004)** 0.0001 (0.00005)* 0.00008 (0.00004)**
N. Obs 26 26 26
R**2 0.79 0.76 0.82
** significant at 5%, ** *significant at 1%

Table 4. Number of pensioners
INDEPENDENT VARIABLES DEPENDENT VARIABLE
NUMPENS (a)
NUMPENS (b)
CONSTANT 4.53 (3.54) -8.72 (1.21)
OLD 1.84 (0.14)*** 1.46 (0.17)***
ACTOT -20.65 (5.95)***
DAC 24.25 (5.13)***
N. Obs 53 53
R**2 0.84 0.87
** *significant at 1%
Table 5. Average pension
INDEPENDENT
VARIABLES
AVPENS
CONSTANT 2.31 (0.62)***
DAC -0.34 (2.52) 1
OLD -0.13 (0.086) 1
N. Obs. 50
R**2 0.1
** *significant at 1% , 1 not significant

τ t+1
Q(τ t
)
τ ss
Figure 1
τ ss
τ t

RELACION DE DOCUMENTOS DE FEDEA
DOCUMENTOS DE TRABAJO
2005-03: “The Evolution of Retirement”, J. Ignacio Conde-Ruiz., Vincenzo Galasso y Paola
Profeta.
2005-02: “Housing deprivation and health status: Evidence from Spain” Luis Ayala, José M.
Labeaga y Carolina Navarro.
2005-01: “¿Qué determina el éxito en unas Oposiciones?”, Manuel F. Bagüés.
2004-29: “Demografía y empleo de los trabajadores próximos a la jubilación en Cataluña”, J. Ignacio
Conde-Ruiz y Emma García
2004-28: “The border effect in Spain”, Salvador Gil-Pareja, Rafael Llorca-Vivero, José A.
Martínez-Serrano y Josep Oliver-Alonso.
2004-27: “Economic Consequences of Widowhood in Europe: Cross-country and Gender Differences”
Namkee Ahn.
2004-26: “Cross-skill Redistribution and the Tradeoff between Unemployment Benefits and
Employment Protection”, Tito Boeri, J. Ignacio Conde-Ruiz y Vincenzo Galasso.
2004-25: “La Antigüedad en el Empleo y los Efectos del Ciclo Económico en los Salarios. El Caso
Argentino”, Ana Carolina Ortega Masagué.
2004-24: “Economic Inequality in Spain: The European Union Household Panel Dataset”, Santiago
Budría y Javier Díaz-Giménez.
2004-23: “Linkages in international stock markets: Evidence from a classification procedure”, Simon
Sosvilla-Rivero y Pedro N. Rodríguez.
2004-22: “Structural Breaks in Volatility: Evidence from the OECD Real Exchange Rates”, Amalia
Morales-Zumaquero y Simon Sosvilla-Rivero.
2004-21: “Endogenous Growth, Capital Utilization and Depreciation”, J. Aznar-Márquez y J. R.
Ruiz-Tamarit.
2004-20: “La política de cohesión europea y la economía española. Evaluación y prospectiva”, Simón
Sosvilla-Rivero y José A. Herce.
2004-19: “Social interactions and the contemporaneous determinants of individuals’ weight”, Joan
Costa-Font y Joan Gil.
2004-18: “Demographic change, immigration, and the labour market: A European perspective”, Juan
F. Jimeno.
2004-17: “The Effect of Immigration on the Employment Opportunities of Native-Born Workers:
Some Evidence for Spain”, Raquel Carrasco, Juan F. Jimeno y Ana Carolina Ortega.
2004-16: “Job Satisfaction in Europe”, Namkee Ahn y Juan Ramón García.
2004-15: “Non-Catastrophic Endogenous Growth and the Environmental Kuznets Curve”, J. Aznar-
Márquez y J. R. Ruiz-Tamarit.
2004-14: “Proyecciones del sistema educativo español ante el boom inmigratorio”, Javier Alonso y
Simón Sosvilla-Rivero.
2004-13: “Millian Efficiency with Endogenous Fertility”, J. Ignacio Conde-Ruiz, Eduardo L.
Giménez y Mikel Pérez-Nievas.
2004-12: “Inflation in open economies with complete markets”, Marco Celentani, J. Ignacio Conde
Ruiz y Klaus Desmet.
TEXTOS EXPRESS
2004-02: “¿Cuán diferentes son las economías europea y americana?”, José A. Herce.
2004-01: “The Spanish economy through the recent slowdown. Current situation and issues for the
immediate future”, José A. Herce y Juan F. Jimeno.